Areas with aerial ac power distribution and frequent lightning strikes such as southern US ( FLorida) impose high failure rates to electronic equipment , even some with MOV protection. The advertised warnings are disconnect your computer during a lightning storm. Even some line filters and MOVs are inadequate in some areas. In the 80's "We" developed methods to survive 3kV, 6kV and 15kV transients and compared the costs for our SMPS in high-rel environments with high-volume, low-cost methods. It is a blend of methods using CM chokes (2), Pi filter, TVS, MOV's and fail-safe plastic caps.
Dysan marketting have no credible evidence to support well-reasoned Dysan engineering concepts, some of which are original for lighting such as the electrical connections used for linear position rods. The heatsinks are old concepts and nothing in their design supports their wild MTBF claims.
Distributed lighting with focal lenses will always be cheaper than centralized LEDs with Cu heat distribution. LED strips are easily fabricated with distributed dissipation with < $0.1/W costs in volume. Microlenses and proper luminaire design do offer putting the light down without the glare of a 10W LED.
The claims that all fluorescent lights are not efficient are blatent lies or ignorance.
I use tri-phosphor tubes with 88 lumens per watt which as good as the majority of LED luminaires and can drive one to four (4) 32W tubes independantly The tri-phosphor gives a better CRI than most single phosphor LEDs with a blue substrate. CRI is the index to render colour accurate on a pallette of pastel test patterns. Although LED's can give better colour saturation on tests, wide spectrum lights render higher CRI for coloured prints but does not mean best contrast for reading, which is the intent of Dysan's luminaires.
Large array LED's with higher power density than a XEON core ( W/mm^2) but can be kept cool better with patented vertical microtubes and another design uses spiral vortex convection fins, better than planar copper vapor phase tubing but effective in tight space like all Laptops for GPU's and some CPU's.
Power meters have a 100kA arc gap set at 6kV built in. After this the best attenuation is by filtering with high insulation rating of 15kV. After that the current is limited to the clipping suppression methods available such as TVS and the rise time is now slow enough for MOV's and has tubes are negative resistance devices which can also fail from follow-on current. The PU and PE caps can absorb the high dv/dt spikes and attenuate further with low ESR ratio from series L*dI/dt, but resonance must still be designed to control maximum Q and resonant frequencies.
The best approach for surge protection is a blend of CM choke rejection in 2 stages, LC Diff mode filtering, thus significantly attenuating a 6kV worst case pulse by filtering, attenuation, clipping, and isolation in the SMPS transformer. Most TV's are only rated to 3kV transients but may survive 5kV or more depending on pulse width.
Dysan's MTBF claims are wild to say the least and based on poor assumptions not engineering evidence. Nowhere do they report the junction rise or actual FIT's, confidence levels used or environmental stress of their life test. It is probably a theoretical comparison to 2.5x 50K theoretical marketing values based on other lamps wild claims of 50khrs but with a 15'C cooler junction temp thus Arhenius improvement of a little more than double the MTBF.
All of this assumes defect free which never exists in a manufacturing process that is not mature.
It must be calculated and must be demonstrated.
For Aerospace, it's a whole different game. They MUST use old technology with big lithography because the gamma rays can significantly reduce breakdown voltage which might start at 10kV/mm or 10V/um and 1um lithography is too small for Aerospace. ( XEON's are 25nm now and pushing 15)